灰色模型GM(1,1)在广西出生缺陷预测中的应用

Effect of gray model(1,1) on prediction of birth defects among the population of Guangxi Zhuang Autonomous Region

目的 应用灰色模型GM(1,1)构建不同水平的出生缺陷发生率预测模型,并对预测模型进行评估,探索灰色模型GM(1,1)在出生缺陷发生率中的预测效果。方法 基于2016-2020年广西出生缺陷监测数据,以年、季度、月共3个层次构建总出生缺陷发生率及前5种出生缺陷的灰色模型GM(1,1),并通过模型综合评价指标对不同层次的灰色模型预测效果进行评估。结果 在使用灰色模型预测总出生缺陷年度、季度、月发生率时,平均相对误差分别为-0.12%、0.49%、0.79%,后验差比值C分别为0.002、0.285、0.392,预测效果分别为好、好、合格;前5种出生缺陷(先天性心脏病、多指(趾)、马蹄内翻足、外耳其他畸形、并指(趾))的年度发生率灰色模型拟合效果明显优于季度和月份的。结论 灰色模型的预测效果可能与原始序列数据波动性有关,在对出生缺陷发生率进行预测时以年度为单位的预测效果可能更佳。

Objective To study the effect of gray model GM(1,1)in predicting birth defects(BD)prevalence at different levels among the population of Guangxi Zhuang Autonomous Region. Methods Based on the monitoring data of BD in Guangxi from 2016 to 2020,the gray mode GM(1,1)was used to predict the overall BD prevalence and the top five main BD at three levels(year,quarter,and month). We compared the prediction effect of different levels of gray models. Results The average relative errors for yearly,quarterly and monthly overall BD prevalence were-0.12%,0.49% and 0.79%,respectively,and the posterior odds ratios were 0.002,0.285and 0.392,respectively,and the prediction results were good,good and qualified,respectively. The yearly gray model prediction effect of the top five main BD(congenital heart disease,polydactyly,talipes equinovarus,other external ear deformities,and syndactyly)was significantly better than those of quarterly and monthly. Conclusion The prediction effect of GM(1,1)may be related to the volatility of original sequence data. It may be better for predicting the BD prevalence by year.